Homework 5 (Computed Tomographic (CT) Reconstruction) : submit via email to ee528homeworks AT gmail.com
I will grade best four out of the five homeworks.

• Take the phantom image
  
• Simulate CT : compute radon transform along N directions and at M points in each direction
• Implement CT reconstruction using two of the following techniques
  
• Filtered Back-Projection (FBP)
  
• FBP with the high pass filter replaced by a Wiener filter  (FBP-Wiener)
  
• Algebraic Reconstruction Technique (ART) with a piecewise constant square/rectangle prior model
  
• Compressive Sensing (CS)  (recreate the results given in this paper http://www.acm.caltech.edu/~jrom/publications/CandesRombergTao_revisedNov2005.pdf )
• Compare the improvement in reconstruction quality as the the number of projection angles, N, is increased from 20 to 90, do this visually for two-three cases and also plot an MSE plot,
  
• If you do implement FBP-Wiener, compare performance with FBP when noise is added to the radon transform
  
• I will recommend implementing either FBP and FBP-Wiener  (compare performance as noise is increased) or FBP and CS (compare performance as N is changed), or ART and CS (compare performance as N is changed)
  
• 
  
• If you want - try the above algorithms on brain or cardiac images obtained from the web